- Fixed a couple of things

- Now the eigenvalue based solutions are nicely consistent with the deSolve solutions, if enough
  output times are specified (100, sometimes more are needed, see test.R)
- Workaround for invilr not to produce NaN values so often
- Still a lot to do (see TODO)



git-svn-id: svn+ssh://svn.r-forge.r-project.org/svnroot/kinfit/pkg/mkin@30 edb9625f-4e0d-4859-8d74-9fd3b1da38cb

5 files changed, 21 insertions, 3 deletions

diff --git a/R/ilr.R b/R/ilr.R
index b53f868..389653e 100644
--- a/R/ilr.R
+++ b/R/ilr.R
@@ -40,5 +40,12 @@ invilr<-function(x) {
   for (i in 1:D) {
     z[i] <- exp(y[i])/sum(exp(y))
   }
+
+  # Work around a numerical problem with NaN values returned by the above
+  # Only works if there is only one NaN value: replace it with 1
+  # if the sum of the other components is < 1e-10
+  if (sum(is.na(z)) == 1 && sum(z[!is.na(z)]) < 1e-10) 
+    z = ifelse(is.na(z), 1, z)
+
   return(z)
 }
diff --git a/R/mkinfit.R b/R/mkinfit.R
index 37eee33..cf58a7d 100644
--- a/R/mkinfit.R
+++ b/R/mkinfit.R
@@ -111,6 +111,9 @@ mkinfit <- function(mkinmod, observed,
   {

     assign("calls", calls+1, inherits=TRUE) # Increase the model solution counter

 

+    # Trace parameter values if quiet is off

+    if(!quiet) cat(P, "\n")

+

     # Time points at which observed data are available

     # Make sure we include time 0, so initial values for state variables are for time 0

     outtimes = sort(unique(c(observed$time,

@@ -305,7 +308,8 @@ mkinfit <- function(mkinmod, observed,
   data$variable <- ordered(data$variable, levels = obs_vars)

   fit$data <- data[order(data$variable, data$time), ]

   fit$atol <- atol

-  fit$parms.all <- parms.all

+  fit$parms.all <- parms.all # Return all backtransformed parameters for summary

+  fit$odeparms.final <- parms.all[mkinmod$parms] # Return ode parameters for further fitting

 

   class(fit) <- c("mkinfit", "modFit") 

   return(fit)

diff --git a/R/mkinmod.R b/R/mkinmod.R
index 54b3b5f..41dbefa 100644
--- a/R/mkinmod.R
+++ b/R/mkinmod.R
@@ -92,7 +92,7 @@ mkinmod <- function(..., use_of_ff = "min")
           parms <- c(parms, k_compound_sink)

           decline_term <- paste(k_compound_sink, "*", box_1)

         } else { # otherwise no decline term needed here

-          decline_term = "" 

+          decline_term = "0" 

         }

       } else {

         k_compound <- paste("k", box_1, sep="_")

@@ -128,6 +128,10 @@ mkinmod <- function(..., use_of_ff = "min")
         reversible_binding_term_2 <- paste("+", k_free_bound, "*", box_1, "-",

           k_bound_free, "*", box_2)

       } else { # Use formation fractions also for the free compartment

+	stop("The maximum use of formation fractions is not supported for SFORB models")

+        # The problems were: Calculation of dissipation times did not work in this case

+        # and the coefficient matrix is not generated correctly by the code present 

+        # in this file in this case

         f_free_bound <- paste("f", varname, "free", "bound", sep="_")

         k_bound_free <- paste("k", varname, "bound", "free", sep="_")

         parms <- c(parms, f_free_bound, k_bound_free)

diff --git a/R/mkinpredict.R b/R/mkinpredict.R
index 49166f3..2b7e51d 100644
--- a/R/mkinpredict.R
+++ b/R/mkinpredict.R
@@ -17,7 +17,9 @@ mkinpredict <- function(mkinmod, odeparms, odeini, outtimes, solution_type = "de
     o <- switch(parent.type,
       SFO = SFO.solution(outtimes, 
           evalparse(parent.name),
-          evalparse(paste("k", parent.name, sep="_"))),
+          ifelse(mkinmod$use_of_ff == "min", 
+	    evalparse(paste("k", parent.name, "sink", sep="_")),
+	    evalparse(paste("k", parent.name, sep="_")))),
       FOMC = FOMC.solution(outtimes,
           evalparse(parent.name),
           evalparse("alpha"), evalparse("beta")),
diff --git a/TODO b/TODO
index 10ccf91..13bc6fc 100644
--- a/TODO
+++ b/TODO
@@ -1,4 +1,5 @@
 - Fix transformation and backtransformation of formation fractions
+- Fix problems with the schaefer complex test data (test.R)
 - Adapt mkinplot function
 - Calculate confidence intervals for parameters assuming normal distribution
 - Calculate confidence intervals for DT50 and DT90 values when only one parameter is involved